Playground stop generating output and have to write continue

Prompting
1

Hi,

I’m using Trubo model in the playground and it always stop generating output and I have to write continue.
How to solve this?

Regards,

2

Ensure that the token limit is at maximum, 2048, and then ask for a “verbose” output, note that the system will not generate output over the maximum token length in one go.

3

Already maximum token is 2048 and the still the same problem

4

If the maximum in one go is 2048, how did they generate 7000 words article?

5

There is no GPT Model that can generate an average of 7000 words, the GPT-4 8k model can generate an average of 6000 words, so until the 32k model becomes public, that is your limit.

6

I have an ai writer works with OpenAI API and it generates 6000 words article with turbo model.
How that works?

7

My guess would be be multiple calls to the API

8

Do you mean for example a call for each heading or so?
But does the output, in this case, will be coherent?

9

Yes, if you craft the prompts carefully and logically you can pass overarching context to all calls and give requests for specialised local information, so you could have a back story maintained across all prompts and then heading/chapter/paragraph prompts appended to that for each call, that way you keep important information logically consistent.